LIGHT AND VEGETATION 51 



precision, or to adopt both courses at the same time. Let us 

 now consider some of these approximate methods. 



We may try to replace the monochromator by a fiher of 

 glass or coloured gelatine, or a vessel containing a coloured 

 Uquid. The spectral bands thus isolated are always rather 

 large, with shaded edges, and infra-red transmission bands, 

 often unsuspected, are particularly misleading (see, for 

 example, the blue filter in Fig. I, 4). It is nearly always necessary 

 to have several filters superposed. 



Also, to avoid the very low Hght efiiciency of the diff'using 

 sphere, several experimenters have used spherical or elliptical 

 poUshed mirrors, of very large aperture, which collect all the 

 rays reflected by the sample placed near the centre, or at a 

 focus, and send them to the conjugate point, which is the 

 other focus in the case of the elUptical mirror. 



The principle appears to be unassailable, but all receiving 

 systems sensitive to light — cells or thermopiles — have a different 

 sensitivity for differently inclined rays; in general, rays with 

 an incidence too far from the perpendicular are much less 

 effectual, so that the more the Hght is diff'used, the less intense 

 it appears with this method, which suff'ers also from other 

 defects. 



On the other hand, the apparatus thus receives much more 

 energy, since the flux reflected by the sample is concentrated 

 on the receiver instead of being dispersed over the whole 

 surface of a diffusing sphere. 



Possible errors are much reduced by comparing the leaf 

 with a white diffuser of magnesia or magnesium carbonate. 



It is, therefore, to avoid the difficulties of measuring very 

 smaU quantities of light that the use of truly monochromatic 

 hght and of a diffusing sphere have been abandoned. 



Even if a really practical solution were found to this 

 purely technical problem, one would still have to remember 

 that the reflection factor depends on the way in which the 

 flux of incident hght is presented. It may be projected in 

 nearly parallel rays, and in this case the angle of incidence 

 may vary from 0° to 90°. It may also be projected in a more 



